基于深度学习的智能交通灯设计
Design of Intelligent Transportation Light Based on Deep Learning
孔令龙 1任仕艳1
作者信息
- 1. 玉林师范学院,广西 玉林 537000
- 折叠
摘要
针对传统交通灯通行时间不灵活,通过对采集交通灯下的车流量信息进行了研究,设计了一种基于深度学习的智能交通灯.首先通过FCN提取车道线,使用聚类算法拟合车道线函数;其次使用以VGG网络作为模型的主干特征提取SSD网络模型,检测车辆的位置信息,将车辆的位置信息与车道线位置信息综合起来,统计当前交通灯下的每个车道上的车流量信息,实验结果表明,基于深度学习的智能交通灯系统实际的应用过程中,车流量统计的精确度为90.69%,基本实现在现实环境的应用.
Abstract
In view of the inflexibility of the passage time in traditional transportation light,an intelligent transportation light based on Deep Learning is designed through research on the traffic flow information collected under the transportation light.Firstly,the lane lines are extracted using FCN,and a clustering algorithm is used to fit the lane line function.Secondly,the SSD network model is extracted,with the VGG network as the backbone feature of model,to detect the positional information of vehicles,and the positional information of vehicles is combined with the positional information of lane lines to count the traffic flow information of each lane under the transportation light.The experimental results show that the intelligent transportation light system based on Deep Learning has an accuracy of 90.69%in the actual application process,which is basically applied in the real environment.
关键词
车辆检测/SSD/FCN/智能交通Key words
vehicle detection/SSD/FCN/intelligent transportation引用本文复制引用
出版年
2024
国家科技期刊平台
NSTL
万方数据